Most founder stories follow a familiar arc. Build, raise, scale, exit. The version that gets less attention but is closer to the median experience in venture-backed startups is the one that ends quietly. There’s still enough runway. The team is small. The product had a demand. And the founder decides, deliberately, that the company is not the right vehicle for the next decade of their life.
Mr. Palash Kala lived that version in the summer of 2025. As co-founder of Hovi AI, he had spent more than a year exploring opportunities at the intersection of AI agents and consumer marketing. He had been admitted to South Park Commons’ competitive Founder Fellowship after an earlier rejection, spent four months in San Francisco at the SPC bootcamp, and built a product for AI-generated playable mobile gaming advertisements. Then he walked away from the million-dollar commitment. The decision drew unusual attention because it ran against the prevailing logic of the AI boom. Money was easy. The category was hot. Mr. Kala chose to step away from the journey anyway, returning what remained of his investors’ capital and writing publicly about the lessons he took from the experience. His September 2025 essay in Tech in Asia has circulated widely in founder and investor circles since.
From Rejection to Fellowship
The path to that decision started, characteristically, with a rejection. As Mr. Kala has written publicly, he and his then co-founder applied to the SPC India membership programme shortly after the firm launched its India operations in July 2024. They were rejected. Seven months later, after pivoting from voice agents into AI for mobile gaming advertising, they applied to the SPC Founder Fellowship. The first interview did not go well. They received a call back. By March 2025, they were in. The fellowship offered four hundred thousand dollars in pre-seed funding for seven per cent of the company, with an additional six hundred thousand dollars guaranteed in any subsequent external round. Within two weeks of accepting, Mr. Kala was in San Francisco at the Game Developers Conference.
“The only thing you have as founders is momentum and energy,” one of the SPC partners told the cohort. The lesson stuck. Mr. Kala has written about the demos every two weeks, the tough-love feedback from SPC partners, and the technical density of the SPC community. He spoke to large gaming companies, including Rovio Entertainment and Scopely, and validated the technical thesis: an AI agent could compress the playable ad creation cycle dramatically.
The Pivot That Did Not Stick
The market was less cooperative. Hovi AI began with voice agents for consumer research and pivoted, after several months, into AI-generated playable advertisements for mobile games. Playable ads are interactive, perform well, and are expensive and slow to produce. The thesis was that an AI agent capable of generating them in hours instead of weeks would unlock significantly more demand from advertisers. The team validated the technical thesis but ran into a market reality.
“We discovered two things at roughly the same time,” Mr. Kala said. “The mobile gaming sector was consolidating, not growing. And faster ad creation alone would not expand the market, because the real bottleneck for advertisers was not making ads. It was the cost of testing them on networks like AppLovin. We could compress a two-week creative cycle into an hour, and the customer was still constrained by what came after.”
The team paused, explored adjacent ideas in marketing AI around Generative Engine Optimization, and found that roughly fifty other startups were already further along. Mr. Kala has called this his second “pivot hell” and has been candid about the decision he reached.
“A successful founder once asked me what desire would make me grind for years, away from family on Sunday nights,” he wrote. “Market opportunity is not enough. You need an irrational desire to build a company. I realised I did not have it for this one.”
What He Learned, and What He Wrote
Several months after stepping away from Hovi, Mr. Kala published the widely shared post laying out the lessons he had taken from the experience. The post argued that the AI boom has made building products painless and made building startups look easy, but that the second is as hard as it ever has been. The post was distributed widely on social media and was featured in Tech in Asia and has been referenced in founder and investor circles in San Francisco and Bangalore in the months since.
His rules of thumb run against the grain of much of the prevailing hype. Start with the audience, not the technology. Find a niche that is too small for incumbents but big enough to grow into. Validate that customers will accept an imperfect first version of your product, because if they need one hundred per cent accuracy from an AI system today, you are probably in the wrong space.
“There is a difference between a product that works and a startup that works,” Mr. Kala said. “AI lets you build the first one quickly. It does not change anything about the second one. Spinning up a prototype in a weekend does not give you a market, a moat, or the conviction to spend a decade on a problem. Early-stage founders who confuse those two things will have an expensive lesson to learn in their journey.”
Building Again, From the Inside
After winding down Hovi, Mr. Kala joined Composio, a San Francisco-based AI agent infrastructure company, in September 2025 as Head of Product. The decision, he has said, was about leverage rather than ambition. After eighteen months of thinking about agents from the application side, working on the infrastructure layer beneath them was a way to apply what he had learned without starting from zero.
He has been clear that he is not done as a founder but that he is in no rush. “I will build another company. I do not know what it will be yet, and I am genuinely fine with that. The mistake at Hovi was not that we picked the wrong idea. It was that I did not have the irrational conviction the journey deserved. I am not making that mistake twice.”
